New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
@wych - nope
Decided to go for the Rackspace offering. Seems they use dns1.stabletransit.com and dns3.stabletransit.com. Are these also used by paying customers? Are there any differences between free and paying customers?
Yes, in fact the product is intended as a value added service for customers which have paid services with them.
TTL is 300. I've moved to Rackspace now and things seem to be okay except it seems they only serve queries from servers in the US.
No, they use anycast, servers in Europe too.
Why so low?
Increase the TTL and let the Internet's DNS caching system do it's job.Try 14400 - 4 hours. If you plan to migrate the domain, reduce the TTL in advance.
For health checks. The DNS entries can be modified by a monitoring server to route traffic to a backup server. We need a low TTL for faster update checks by caching servers.
> sleddog said: Why so low?For health checks. The DNS entries can be modified by a monitoring server to route traffic to a backup server. We need a low TTL for faster update checks by caching servers.EDIT: This is called a double post due to BSOD due to horrible Nvidia drivers.
...which many DNS servers are going to ignore anyway. Lots will not honor a 300 second TTL. 2 hours minimum with some of the big guys.
Either
This is against the RFC standards. Although some of the
greedyDNS providers will not conform with these standards. Although, I optimistic hope for humanity.I have seen conflicting reports of the ttl of the largest DNS servers. Some report Google's to cache at a minimum of 2 seconds, others report a minimum of 24 hours. Some say that OpenDNS will cache for 0 seconds, while others say 1 hour. I hope the majority will follow standards, and we should develop against standards - since this is what standards were made for.
RRDNS only works when the backup servers are functional 100% of the time. It may not be optimal for some service providers.
HA is used after the client knows where to route data. What if the load balancer goes down?
I believe that DNS level routing is the best solution to handle the case of whole data center downtime. I don't know of any other solutions.
cloudflare! i notice when you modify A record in cloudflare it resolves very fast. not sure why. maybe because of my location?
Welcome to reality.
No. Current browsers do a RRDNS lookup and then will try servers until they find one that works. If one fails, they go on to the next one.
This is client behavior but is supported in current browsers.
You need to understand a virtual IP. But you won't find it at LEB providers.
Or have multiple load balancers...
BGP.
Your solution really does not work. It may work in most cases, but it's wasteful. A solution that provides HA in "most" situations is not really HA.
Dyn dns
I'm sure OP has already chosen but I'd definitely go DNS4.pro. Free Anycast.